Ontology-Based Data Stream Selection for Real-Time Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing systems, such as stream reasoning, face challenges in flexibility and real-time performance due to increased data processing amounts during inference operations.
Innovation Solution
A data processing system that separates processing into inference and raw data stages, using a storing unit for ontology data, a selection unit to infer key information, and a processing unit to abstract and structure data streams, allowing for flexible data processing and improved speed by selecting and processing data streams using ontology data and machine learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If forward inference or backward inference is performed on RAW data using stream reasoning, then flexibility in data processing is improved by allocating multiple ontologies to single data, but the amount of data processing increases and real time performance deteriorates
Solution Approach 1:
The patent pre-generates inference results for multiple ontologies and stores them in advance before actual data processing. When RAW data arrives, the system directly retrieves pre-computed inference results instead of performing inference operations in real-time, thus achieving both flexibility (multiple ontologies) and real-time performance (no computation delay).
Solution Approach 2:
The patent separates the inference process from the data processing process. Inference operations are performed independently in advance and stored as separate results, while the main data processing flow directly uses these pre-computed results. This segmentation eliminates the computational burden from the real-time data processing path.
2Productivity
If stream data middleware or complex event processing is used to perform real time processing on data, then processing speed is improved, but flexibility deteriorates because a single context is allocated to a single piece of data
Solution Approach 1:
The patent introduces pre-computed inference results as an intermediary layer between RAW data and processing operations. This intermediary contains pre-analyzed information from multiple ontologies, allowing the system to maintain fast processing speeds while accessing rich contextual information that would otherwise require complex real-time inference.
Solution Approach 2:
The system performs context analysis and ontology inference in advance, storing results that can be quickly retrieved during processing. This preliminary preparation enables fast processing without sacrificing the ability to apply multiple ontological contexts to each data item.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
There is provided a data processing system. A storing unit stores ontology data. A selection unit selects a data stream including key information corresponding to a query using the ontology data stored in the storing unit, among data streams including key information uniquely given to sensor data. A processing unit processes the selected data stream.